system-architecture-translator
Skill1 starsChanged 4 months agoMITView on GitHub
Translate data flows, services, queues, workflows, storage, consistency, latency, failures, retries, and distributed-system constraints into visible product and UX decisions for a case study. Use for AI agents, desktop tools, data products, scraping engines, background jobs, cloud systems, or any portfolio story where architecture determines telemetry, state management, loading, recovery, permissions, or user control.
In this repository11 other files it gives its agents.
The file
Files
System Architecture Translator
Explain how invisible system behavior shaped the interface. Read references/system-to-ux-map.md.
Workflow
- Inspect architecture diagrams, code paths, API contracts, schemas, traces, logs, incidents, and product states.
- Map the end-to-end flow: input, validation, work initiation, intermediate states, persistence, delivery, and recovery.
- Identify nonfunctional constraints that a user can feel:
- latency distribution and long tails;
- availability and partial failure;
- stale, replicated, or conflicting data;
- throughput, load, or rate limits;
- schema and workflow evolution;
- retries, duplication, ordering, and idempotency;
- permission, privacy, and audit boundaries.
- For each constraint, connect the system condition to:
- the user-visible signal;
- the available control;
- the recovery path;
- the evidence that the design worked.
- Separate operational telemetry from user-facing explanation.
- Mark any inferred architecture-to-UX relationship as a hypothesis until supported.
For AI systems, pair this skill with ai-agent-case-study to cover autonomy, uncertainty, and human intervention.
Required sections
Return:
- a system boundary and data-flow summary;
- a System→UX translation map;
- state, latency, and progress behavior;
- failure, retry, rollback, and recovery behavior;
- telemetry and audit design;
- architectural trade-offs and unresolved evidence.
Example
Input:
Jobs fan out across three APIs. Completion ranges from 8 seconds to 4 minutes.
Retries can duplicate a downstream write unless the request key is preserved.Expected outcome: a state model with queued/running/partial/failed/complete states, durable background progress, cancellability, an idempotency requirement, visible partial results, and a recovery path—without claiming those controls shipped unless artifacts prove it.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

